Predictive model of playing position in soccer players u-20 150 through machine learning
DOI:
https://doi.org/10.47197/retos.v83.118967Keywords:
Machine Learning, Playing Position, Predictive Model, SoccerAbstract
Introduction: Identifying playing positions through physical variables may contribute to optimizing training planning in young soccer players by supporting objective decision-making during talent identification and individualized training prescription. Objective: The aim of this study was to develop a supervised multiclass classification model to predict playing position in under-20 soccer players using anthropometric and physical performance variables.
Methodology: A quantitative, observational, cross-sectional, and predictive study was conducted with 160 under-20 soccer players equally distributed among goalkeepers, defenders, midfielders, and forwards. Anthropometric variables, speed, agility, strength, power, aerobic endurance, and training load were assessed using standardized field-based tests. Five supervised classification algorithms were compared, and their performance was evaluated using stratified cross-validation and standard classification metrics, including accuracy, precision, recall, and macro F1-score.
Results: The Gaussian Naïve Bayes model achieved the best performance (macro F1-score = 0.728), reaching an overall accuracy of 78% on the independent test set. The variables with the greatest predictive importance were the 30-m sprint, COD505 change-of-direction test, Agility T-Test, 20-m sprint, Yo-Yo test, and VO₂max, whereas anthropometric variables contributed less to the prediction.
Conclusions: Speed, agility, and aerobic capacity were the main predictors of playing position in under-20 soccer players. Machine learning models represent an objective tool for identifying positional profiles, supporting individualized training programs, and providing useful information for player assessment and long-term athletic development.
References
Alpaydin, E. (2020). Introduction to machine learning (4th ed.). MIT Press.
Azcárate Jiménez, U., & Yanci Irigoyen, J. (2016). Perfil físico en futbolistas de categoría amateur de acuerdo a la posición que ocupan en el campo. Revista Española de Educación Física y Deportes, (415), 21–37. https://doi.org/10.55166/reefd.v0i415.504
Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32. https://doi.org/10.1023/A:1010933404324
Çetin, O., & Koçak, M. (2022). Repeated sprint ability of youth football players in the same age category according to playing position and competition level. Montenegrin Journal of Sports Science and Medicine, 11(1), 59–63. https://doi.org/10.26773/mjssm.220307
Cometti, G. (2007). La preparación física en el fútbol. Editorial Paidotribo.
Efron, B., & Tibshirani, R. J. (1993). An introduction to the bootstrap. Chapman & Hall/CRC. https://doi.org/10.1201/9780429246593
Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters, 27(8), 861–874. https://doi.org/10.1016/j.patrec.2005.10.010
Fisher, A., Rudin, C., & Dominici, F. (2019). All Models are Wrong, but Many are Useful: Learning a Vari-able's Importance by Studying an Entire Class of Prediction Models Simultaneously. Journal of machine learning research : JMLR, 20, 177.
Harris, C. R., Millman, K. J., van der Walt, S. J., et al. (2020). Array programming with NumPy. Nature, 585(7825), 357–362. https://doi.org/10.1038/s41586-020-2649-2
Horníková, H., & Zemková, E. (2021). Relationship between physical factors and change of direction speed in team sports. Applied Sciences, 11(2), 655. https://doi.org/10.3390/app11020655
Jiménez-Reyes, P., Cuadrado-Peñafiel, V., & González-Badillo, J. J. (2011). Análisis de variables medidas en salto vertical relacionadas con el rendimiento deportivo y su aplicación al entrenamiento. Cultura, Ciencia y Deporte, 6(17), 113–119. https://doi.org/10.12800/ccd.v6i17.38
Keiner, M., Kapsecker, A., Stefer, T., Kadlubowski, B., & Wirth, K. (2021). Differences in squat jump, line-ar sprint, and change-of-direction performance among youth soccer players according to com-petitive level. Sports, 9(11), 149. https://doi.org/10.3390/sports9110149
McKinney, W. (2010). Data structures for statistical computing in Python. Proceedings of the 9th Py-thon in Science Conference, 56–61. https://doi.org/10.25080/Majora-92bf1922-00a
Medina-Curimilma, W. E. (2025). La importancia de la antropometría en el rendimiento deportivo del fútbol. MQRInvestigar, 9(1), e58. https://doi.org/10.56048/MQR20225.9.1.2025.e58
Moya, D., Tipantuña, C., Villa, G., Calderón-Hinojosa, X., Rivadeneira, B., & Álvarez, R. (2025). Machine learning applied to professional football: Performance improvement and results prediction. Machine Learning and Knowledge Extraction, 7(3), 85. https://doi.org/10.3390/make7030085
Musham, C., & Fitzpatrick, J. F. (2020). Familiarisation and reliability of the isometric mid-thigh pull in elite youth soccer players. Sport Performance & Science Reports, 85, 1–4.
Pedregosa, F., Varoquaux, G., Gramfort, A., et al. (2011). Scikit-learn: Machine learning in Python. Jour-nal of Machine Learning Research, 12, 2825–2830.
Raya, M. A., Gailey, R. S., Gaunaurd, I. A., Jayne, D. M., Campbell, S. M., Gagne, E., Manrique, P. G., Muller, D. G., & Tucker, C. (2013). Comparison of three agility tests with male servicemembers: Edgren Side Step Test, T-Test, and Illinois Agility Test. Journal of Rehabilitation Research and Develop-ment, 50(7), 951–960. https://doi.org/10.1682/JRRD.2012.05.0096
Sammoud, S., Negra, Y., Bouguezzi, R., Ramirez-Campillo, R., Moran, J., Bishop, C., & Chaabene, H. (2024). Effects of plyometric jump training on measures of physical fitness and lower-limb asymme-tries in prepubertal male soccer players: A randomized controlled trial. BMC Sports Science, Medicine and Rehabilitation, 16, 37. https://doi.org/10.1186/s13102-024-00821-9
Santacruz Marín, C. A., España Chacua, W. A., Gaviria Chavarro, J., Motato Rodríguez, L. A., & Galeano Virgen, J. D. (2025). Modelo predictivo del sprint basado en salto contramovimiento e índice de masa corporal en futbolistas. Sportis Scientific Journal, 11(4), 1–19. https://doi.org/10.17979/sportis.2025.11.4.11646
Sneha, S., Prithvi, B. S., Niranjanamurthy, M., Kiran Kumar, H. K., & Dayananda, P. (2024). Machine learn-ing based assessment of elite football players based on anthropometric and motor fitness pa-rameters with regard to their playing positions. SN Computer Science, 5, 974. https://doi.org/10.1007/s42979-024-03261-x
Sokolova, M., & Lapalme, G. (2009). A systematic analysis of performance measures for classification tasks. Information Processing & Management, 45(4), 427–437. https://doi.org/10.1016/j.ipm.2009.03.002
Sporiš, G., Jukić, I., Ostojic, S. M., & Milanović, D. (2009). Fitness profiling in soccer: Physical and physio-logic characteristics of elite players. Journal of Strength and Conditioning Research, 23(7), 1947–1953. https://doi.org/10.1519/JSC.0b013e3181b3e141
Vidal-Maturana, F., Gajardo-Cid, N., Merino-Muñoz, P., Hermosilla-Palma, F., Valdés-Badilla, P., & Herre-ra-Valenzuela, T. (2024). Relación entre potencia y fuerza isométrica con la capacidad de repe-tir sprint en futbolistas profesionales varones. SPORT TK-Revista Euroamericana de Ciencias del Deporte, 13(Supl. 3), 8. 10.6018/sportk.634671
Yagin, F. H., Hasan, U. C., Clemente, F. M., Eken, O., Badicu, G., & Gulu, M. (2023). Using machine learning to determine the positions of professional soccer players in terms of biomechanical variables. Proceedings of the Institution of Mechanical Engineers, Part P: Journal of Sports Engineering and Technology. Advance online publication. https://doi.org/10.1177/17543371231199814
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Javier Gaviria Chavarro, Óscar Hernán Jiménez Trujillo, Nelson Daniel Marin, Juan José Gómez Peñaranda, Francisco Castro Valencia, Miguel Ángel Gómez García, Carlos Julio Grisales Guerra
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and ensure the magazine the right to be the first publication of the work as licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgment of authorship of the work and the initial publication in this magazine.
- Authors can establish separate additional agreements for non-exclusive distribution of the version of the work published in the journal (eg, to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Is allowed and authors are encouraged to disseminate their work electronically (eg, in institutional repositories or on their own website) prior to and during the submission process, as it can lead to productive exchanges, as well as to a subpoena more Early and more of published work (See The Effect of Open Access) (in English).
This journal provides immediate open access to its content (BOAI, http://legacy.earlham.edu/~peters/fos/boaifaq.htm#openaccess) on the principle that making research freely available to the public supports a greater global exchange of knowledge. The authors may download the papers from the journal website, or will be provided with the PDF version of the article via e-mail.
